Interactive education platform based on virtual reality technology

Abstract

The invention relates to the technical field of virtual reality, and discloses an interactive education platform based on a virtual reality technology, which comprises an acquisition and preprocessing module, a scene generation and interaction module, a self-adaptive learning module, an interaction module, an evaluation module and an evidence storage module, by constructing a dynamic interactive virtual scene and a multi-modal feedback mechanism, the problems of scene staticization and single interaction mode in traditional virtual reality education are solved, real-time disassembly, combination and parameter adjustment of scene elements are realized by means of a UE5 engine and a programmed generation algorithm, and eye movement tracking, gesture recognition and tactile feedback technologies are combined, so that the real-time interaction of the scene elements is realized. The immersion and participation degree of the learning process are significantly improved, and students can deepen the understanding and application of knowledge in immersive operation.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of virtual reality, and more particularly discloses an interactive education platform based on virtual reality technology. BACKGROUND

[0002] Virtual reality is a three-dimensional dynamic simulation system that generates a realistic virtual environment through computer technology and allows users to immerse themselves in it and interact in real time through sensory interaction devices.

[0003] The prior art patent document with the authorization announcement number CN116205598A discloses an online education platform based on virtual reality technology, which includes a platform control center, the platform control center is communicatively connected with a teacher module, a student module, a virtual image construction module and a course evaluation module, wherein the teacher module is composed of a courseware design unit, a course unit and an acquisition unit, the courseware design unit is used for the teacher to design a virtual reality courseware, the course unit is used for managing students, and the acquisition unit is used for acquiring the motion trajectory record of the teacher; the student module is composed of a course unit, a problem feedback unit and an acquisition unit; the virtual image construction module is used for constructing the virtual image of the teacher and the student; the course evaluation module is used for supervising the classroom behavior of the student and evaluating the learning quality of the student.

[0004] And the patent document with the authorization announcement number CN118505461A discloses a digital education optimization method and system based on virtual reality technology, which includes collecting learning progress information and learning habit information of students in different learning modules in a virtual reality environment, comprehensively analyzing the learning progress information and learning habit information of students in learning modules, obtaining the ranking of learning performance of students in different learning modules, monitoring learning behavior information and interaction information in learning modules with poor learning performance of students, and comprehensively analyzing the learning behavior information and participation information of students to determine the reasons for poor learning performance of students in learning modules.

[0005] Although the prior art can simplify courseware editing by integrating teaching resources in a database, realize two-way supervision of teachers and students, and identify learning difficulties and adjust learning difficulties and attention problems by analyzing learning module performance, it provides a certain basis for virtual reality education; however, the evaluation dimension is limited to static data such as posture and homework errors, lacks depth analysis of dynamic interactive behavior, and the virtual scene is mostly preset static content, the interaction mode is single, and the feedback adjustment depends on fixed thresholds, lacks personalized adaptation; emotional interaction is missing, and it is difficult to stimulate learning resonance. SUMMARY

[0006] The technical problem solved by the present application is to provide an interactive education platform based on virtual reality technology, which can solve the problems raised in the background art.

[0007] To solve the above technical problems, according to one aspect of the present application, more specifically, an interactive education platform based on virtual reality technology, comprising: the acquisition and preprocessing module is used for acquiring multi-source interaction data and cleaning and standardizing, providing high-quality basic data input for the system, ensuring the reliability of subsequent module analysis, solving the problem of poor data quality and non-uniform format affecting the interaction effect in traditional technology; the scene generation and interaction module is used for constructing dynamic virtual scene and realizing multi-modal interaction feedback, providing an immersive interactive environment for users, solving the problem of static scene and single interaction mode in traditional technology; the adaptive learning module is used for collecting user interaction data, analyzing learning state and dynamically adjusting scene content, realizing personalized learning adaptation, solving the problem of fixed content and lack of targeted adjustment in traditional technology; the interaction module is used for capturing user facial features, calculating emotional state and driving virtual characters to respond, enhancing emotional resonance, solving the problem of lack of emotional interaction in the interaction process; the evaluation module is used for collecting learning data from behavior, cognition and emotion, and then converting it into radar chart and fluctuation curve through visual processing, real-time feedback of learning situation, solving the problem of single evaluation dimension and feedback lag in traditional technology, providing basis for bidirectional optimization of teaching and learning; the storage module generates an unforgeable hash value for the key learning nodes of virtual project achievements and innovative operations, and at the same time, builds a permission management mechanism, which is controlled by the user to control data access permission, supports authorized tracking of learning track, solves the problem of easy tampering of learning data and difficult authentication, and provides a reliable basis for admission and ability authentication.

[0008] Further, the acquisition and preprocessing module comprises: an acquisition module and a preprocessing module.

[0009] Acquisition module: through VR head-mounted sensor, handle sensor and camera, user physiological data, operation behavior data and environmental interaction data are collected.

[0010] Preprocessing module: filter noise in collected data, standardize effective data format, and output structured data that can be directly used for subsequent module analysis.

[0011] Further, the scene generation and interaction module comprises: a scene generation module and a feedback module.

[0012] Scene generation module: based on UE5 engine combined with programmatic generation algorithm, according to teaching target, model library is called, interactive virtual scene is dynamically constructed, user can disassemble, combine and adjust parameters of scene elements;

[0013] Feedback module: integrate multi-modal feedback logic, including eye tracking, gesture recognition, tactile feedback, real-time feedback, enhance the immersion and guidance of interaction.

[0014] Further, the adaptive learning module comprises an interaction collection module, an analysis module and an adjustment module.

[0015] The interaction collection module collects interaction data of the user in the virtual scene in real time, including operation duration, error type, path selection and repeated operation times.

[0016] The analysis module models the interaction data by using an LSTM neural network, identifies a user learning state label, and outputs a state feature vector.

[0017] The adjustment module dynamically adjusts scene content according to the learning state feature vector.

[0018] Further, the interaction module comprises an expression capture module, a response module and an emotion calculation module.

[0019] The expression capture module collects user facial muscle movement data through a camera, extracts key feature point coordinates of eyebrow arches and mouth corners, and generates an expression feature sequence.

[0020] The response module drives a virtual character to make a response according to a dominant emotion type, including expression simulation, voice adjustment and action output.

[0021] The emotion calculation module matches the expression feature sequence with a preset emotion label library, calculates an emotion state confidence, and outputs a dominant emotion type.

[0022] Further, the evaluation module comprises a data collection module and a visualization module.

[0023] The data collection module collects learning data from multiple dimensions, including behavior data, cognitive data and emotion data.

[0024] The visualization module converts multi-dimensional data into intuitive charts, wherein a radar chart is used to show score distribution of behavior, cognition and emotion dimensions, and a fluctuation curve is used to present real-time change trends of the multi-dimensional data in the learning process.

[0025] Further, the storage module comprises a key node module and an access management module.

[0026] The key node module generates an unalterable hash value for key learning nodes of virtual project achievements, innovative operations and evaluation results, and stores the hash value in a blockchain after associating with a time stamp.

[0027] The access management module sets a data access permission level by the user, supports an authorized object to trace original data of a corresponding learning track through a hash value, and ensures data authenticity and controllability.

[0028] The interactive education platform based on the virtual reality technology has the following advantages:

[0029] By constructing dynamic and interactive virtual scenes and multimodal feedback mechanisms, the problems of static scenes and single interaction methods in traditional virtual reality education have been solved. With the help of the UE5 engine and procedural generation algorithms, the scene elements can be disassembled, combined and their parameters adjusted in real time. Combined with eye tracking, gesture recognition and haptic feedback technologies, the immersion and participation in the learning process are significantly improved, allowing students to deepen their understanding and application of knowledge through immersive operation.

[0030] By integrating adaptive learning, emotional interaction, multi-dimensional assessment, and blockchain-based evidence storage modules, personalized learning paths and credible learning certification are achieved. The dynamic analysis of learning status by the LSTM neural network ensures the targeted adjustment of content, while the virtual character response driven by emotion computing enhances learning resonance. Multi-dimensional visual assessment and tamper-proof learning data evidence storage provide teachers and students with real-time optimization basis and credible support for further education and competency certification, comprehensively improving the practicality and credibility of virtual reality education. Attached Figure Description

[0031] The present invention will now be described in further detail with reference to the accompanying drawings and specific implementation methods.

[0032] Fig. 1 This is a schematic diagram of the system principle;

[0033] Fig. 2 This is a flowchart illustrating the steps. Detailed Implementation

[0034] The present invention will be described in detail below with reference to the accompanying drawings and embodiments. It should be noted that, unless otherwise specified, the embodiments and features described in the present application can be combined with each other.

[0035] According to one aspect of the invention, such as Figs. 1-2 As shown, an interactive education platform based on virtual reality technology is provided, including: a data acquisition and preprocessing module, used to acquire multi-source interactive data and perform cleaning and standardization, providing high-quality basic data input for the system, ensuring the reliability of subsequent module analysis, and solving the problems of poor data quality and inconsistent formats affecting the interactive effect in traditional technologies. This module includes:

[0036] Data Acquisition Module: Collects user physiological data, operational behavior data, and environmental interaction data through VR headset sensors, controller sensors, and cameras;

[0037] The VR headset has an infrared eye movement sensor to capture the pupil movement trajectory, a handle sensor to record the position and attitude changes of the handle six-axis movement sensor, and a camera for RGB-D camera to collect depth information to assist behavior analysis. The physiological data such as eye movement trajectory and facial muscle movement, the operation behavior data such as handle button timing and gesture posture, and the environmental interaction data such as virtual object collision and scene element clicking are accurately obtained.

[0038] The preprocessing module filters the noise in the collected data, standardizes the effective data format, and outputs structured data that can be directly used for subsequent module analysis.

[0039] The collected data is filtered by using Kalman filtering algorithm. The algorithm establishes the state equation of the dynamic system as follows:

[0040] x k =Ax k-1 +w k-1

[0041] and the observation equation:

[0042] zX=Hx k +v k

[0043] where x k is the system state at time k, A is the state transition matrix, w k-1 is the process noise, z k is the observation value, H is the observation matrix, and υ k is the observation noise. The optimal estimate at the current time is recursively calculated by combining the estimated value at the previous time and the observation value at the current time. It can effectively filter the handle drift data, random jitter in eye movement tracking, and measurement errors of RGB-D camera depth data, and other sensor noises. At the same time, through data cleaning, instantaneous clicking and other misoperation records less than 0.5 seconds are removed. Such instantaneous operations are usually caused by hardware mis-touch or user's unconscious actions, and do not have actual interactive significance. After removing, the invalid data can reduce the interference of subsequent analysis;

[0044] Format standardization involves converting valid data into JSON format. The specific structure includes "timestamp" (accurate to milliseconds), "data_type" (data type, such as "eye_tracking" or "gesture"), and "values" (data values, which store specific parameters in the form of arrays or key-value pairs; for example, eye-tracking data includes gaze coordinates (x, y), and gesture data includes joint angle values). The output is structured data containing timestamps, data types, and values, ensuring that subsequent modules can directly call it. The standardized data format is unified, which facilitates rapid parsing and use by different components such as the scene generation module and the adaptive learning module, improving the overall data flow efficiency of the system.

[0045] The scene generation and interaction module constructs dynamic virtual scenes and implements multimodal interactive feedback, providing users with an immersive interactive environment and solving the problems of static scenes and limited interaction methods in traditional scenarios. This module includes:

[0046] Scene generation module: Based on the UE5 engine and combined with a procedural generation algorithm, it calls the model library according to the teaching objectives to dynamically construct interactive virtual scenes, supporting users to disassemble, combine, and adjust parameters of scene elements;

[0047] The UE5 engine uses the Nanite virtual geometry system to render massive polygonal models in real time, ensuring high-definition display of complex structures in virtual scenes (such as microscopic cell models and details of ancient buildings). At the same time, the Lumen global illumination system dynamically calculates light reflection and refraction, making the lighting effects of the virtual environment closer to the real world and enhancing the sense of immersion.

[0048] The procedural generation algorithms include:

[0049] The Perlin noise algorithm, used to generate terrain height fields, is expressed as follows:

[0050]

[0051] Where noise() is the basic noise function, and f(x) is the output value of the noise function. is the amplitude scaling factor of the i-th layer noise, which simulates natural terrain undulations through multi-scale noise superposition;

[0052] The L-system algorithm generates a virtual plant model with fractal characteristics by defining an initial string (such as "F" representing a stem segment) and iteration rules (such as "F→FF+[+FFF]-[-F+F+F]"). It allows users to change the plant morphology by adjusting parameters (such as the number of iterations and the branch angle), making it suitable for demonstrating plant growth patterns in biology classes.

[0053] By combining the UE5 with procedural generation algorithms, the scene generation module can complete the construction of complex scenes of polygons and support users to modify physical parameters such as gravity and friction coefficient in real time and observe the dynamic response of scene elements (such as the change in the falling speed of objects after adjusting the gravity).

[0054] The feedback module integrates multi-modal feedback logic, including eye tracking, gesture recognition, and tactile feedback, to provide real-time feedback and enhance the immersion and guidance of the interaction.

[0055] The eye tracking uses the corneal reflection method to locate the pupil center and corneal reflection point and calculates the gaze point coordinates (x, y). When the gaze point deviates from the current learning task area (such as the formula area of the virtual blackboard) for more than 3 seconds, the scene highlights the feedback (the edge of the area flashes a blue halo), and at the same time, a prompt sound is emitted from the corresponding direction through spatial audio technology (such as "Please pay attention to this knowledge point").

[0056] The gesture recognition uses a skeletal key point matching algorithm to calculate the three-dimensional coordinates of the 21 joints of the hand in real time. When the angle between the thumb and index finger is detected to be <30°, it is determined to be a "grabbing" gesture, which is mapped to the picking operation of the object in the virtual scene. If the virtual beaker is not aligned with the bottle opening when grabbing, the handle will give an error prompt through 100Hz high-frequency vibration.

[0057] In addition, the tactile feedback is based on the pulse width modulation (PWM) technology, which simulates different touch sensations by adjusting the pulse duty cycle of the vibration motor: strong vibration with 50% duty cycle when assembling metal parts, weak vibration with 20% duty cycle when touching cloth, and pulse vibration with 200ms interval to simulate the touch sensation of electric current "flowing" when the student completes the correct connection of the virtual circuit, strengthening the sense of operation achievement.

[0058] The adaptive learning module collects user interaction data, analyzes learning status, and dynamically adjusts scene content to achieve personalized learning adaptation and solve the problem of traditional content fixation and lack of targeted adjustment. This module includes:

[0059] The interaction collection module collects user interaction data in the virtual scene in real time, including operation duration, error type, path selection, and repeated operation times.

[0060] The edge computing technology is used to complete data collection and preliminary processing locally on the VR device to avoid cloud transmission delay. The sampling frequency is 10Hz (data is recorded once every 0.1 second) to ensure that details of fast operations (such as reagent mixing actions in virtual experiments) can be captured.

[0061] The collected data is time-stamped and associated with the current learning node (such as "Physics Experiment - Buoyancy Measurement") to facilitate subsequent analysis of the learning performance of this node.

[0062] Analysis module: LSTM neural network is used to model interactive data, identify user learning state labels, and output state feature vectors;

[0063] Wherein the LSTM neural network processes time series data through three hidden layers (64 neurons per layer), the input is a vector containing 16 features such as operation duration standard deviation, error type distribution, path jump frequency, and the output is a 128-dimensional state feature vector. The vector is matched with the preset learning state label library (such as "operation proficiency" and "weak spatial imagination") by cosine similarity calculation.

[0064] Adjustment module: dynamically adjust scene content according to learning state feature vectors;

[0065] Wherein the basic adjustment logic is realized by rule engine, for example, when the feature vector is "the number of repeated errors in the same step > 5", the step is automatically split into 3 sub-steps, each sub-step is equipped with animation demonstration;

[0066] And the complex scene adjustment adopts case reasoning mechanism, retrieves the case with the highest similarity with the current feature vector in the historical database, and reuses the verified adjustment scheme (such as calling 3D model step-by-step rotation animation for "weak spatial imagination" students), while supporting teachers to manually add adjustment rules (such as "when the error rate of a certain knowledge point > 30%, push additional examples"), realizing human-computer collaborative optimization.

[0067] Through the above cooperation, a closed loop of "real-time collection-accurate analysis-dynamic adaptation" is formed, which not only guarantees the timeliness and accuracy of data analysis, but also realizes the targeted adjustment of content through human-computer collaborative optimization, effectively solving the problems of fixed content and feedback lag in traditional education, and significantly improving the learning adaptability and efficiency.

[0068] Interaction module, used to capture user expression features, calculate emotional state and drive virtual characters to respond, enhance emotional resonance and solve the problem of lack of emotional interaction in the interaction process. The module includes:

[0069] Expression capture module: capture user facial muscle movement data through camera, extract key feature point coordinates of eyebrow arch and mouth corner, and generate expression feature sequence;

[0070] Wherein AAM algorithm is used to fit the collected facial image, accurately extract three-dimensional coordinates of key areas such as eyebrow arch (10 feature points) and mouth corner (8 feature points), and generate 30 frames of expression feature sequence per second. The sequence contains the displacement change amount of each feature point (such as eyebrow arch lifting distance and mouth corner pulling amplitude), which provides fine-grained data for subsequent emotional analysis.

[0071] Response module: according to the dominant emotion type, the virtual character makes a response, including expression simulation, voice adjustment and action output;

[0072] Among them, expression simulation adopts BlendShape technology, which realizes expression change by adjusting the vertex weight of the virtual character's face: when "confusion" emotion is detected, the eyebrow arch vertex weight increases (simulating lifting), and the mouth corner vertex weight decreases (simulating pulling down);

[0073] And voice adjustment is based on TTS (text-to-speech) technology, which changes voice parameters according to emotion type: "joy" increases the fundamental frequency and speeds up the speech, and "depression" decreases the fundamental frequency and slows down the speech;

[0074] In addition, action output calls preset animation clips, such as "encouragement" corresponding to the combination of waving hands and nodding, and "doubt" corresponding to tilting head and frowning, to ensure that the virtual character's response matches the emotion type.

[0075] Emotion calculation module: match the expression feature sequence with the preset emotion label library, calculate the emotion state confidence, and output the dominant emotion type;

[0076] Among them, the dynamic time warping (DTW) algorithm is adopted to calculate the similarity between the input expression feature sequence and the standard sequence (such as the feature template of "confusion") in the label library, and the similarity is converted to a confidence in the interval [0, 1] by the Softmax function. The preset label library contains several basic emotions such as "confusion, joy, depression, and concentration". When the confidence of a certain emotion is greater than 60%, it is determined as the dominant emotion type. If the highest confidence is less than 60%, output "neutral" state to avoid misjudgment.

[0077] The three work together, and the delay from expression collection to virtual character response can be controlled within 500ms, making the emotional interaction more real-time and immersive. For example, when the student has a frown (confusion) because of difficulty in solving the problem, the virtual teacher can quickly and synchronously make a concerned expression and slow down the explanation speed.

[0078] The evaluation module is used to collect learning data from behavior, cognition and emotion, and then convert it into radar chart and fluctuation curve through visualization processing, to provide basis for teaching and learning optimization. The module includes:

[0079] Data collection module: collect learning data from multiple dimensions, including behavior data, cognitive data and emotional data;

[0080] The collected data will be associated with specific knowledge points and learning stages (such as "Mathematics - Solid Geometry - Sphere Volume Calculation") and labeled with class average levels (such as a class average error rate of 25% for this knowledge point). When individual data deviates from the average level by more than 30%, it will be marked as "need attention".

[0081] The data collection range covers pre-class preparation, in-class operation, and post-class practice, ensuring comprehensive evaluation of the learning cycle. For example, if the concept understanding error rate is greater than 40% during pre-class preparation, it will trigger a warning for in-class emphasis.

[0082] Visualization module: converts multi-dimensional data into intuitive charts, with radar charts showing behavior, cognition, and emotional dimension score distribution, and fluctuation curves presenting real-time changes in each dimension during the learning process.

[0083] The radar chart is drawn using the D3.js library and includes behavior specification, cognitive load, emotional involvement, and other dimensions. Each dimension has a maximum score of 10 points. The color gradient (green to red) shows the score level, making it easy to quickly identify weaknesses.

[0084] The fluctuation curve is superimposed with a moving average curve (window size of 5 data points) to smooth out transient fluctuations and highlight trend changes (such as an error rate of 3 minutes from 10% to 40% for a knowledge point). Clicking on a point on the curve will display the corresponding analysis (such as "error reason: not considering air resistance"). The chart update frequency is synchronized with the learning progress to ensure timely feedback.

[0085] Proof module: generates non-tamperable hash value proof for virtual project achievements and key learning nodes of innovative operations. It also builds a permission management mechanism, allowing users to control data access permissions, supports authorized tracking of learning tracks, and solves the problem of easy tampering and difficult authentication of learning data, providing reliable evidence for admission and ability certification. This module includes:

[0086] Key node module: generates non-tamperable hash values for virtual project achievements, innovative operations, and evaluation results, and stores them in the blockchain after associating with timestamps.

[0087] The hash value calculation uses the SHA-256 algorithm, with the formula:

[0088] H = SHA-256(data + timestamp + nodeID)

[0089] Wherein H is the final output hash value, data is the original data to be stored, timestamp is the timestamp, nodeID is the node identification, ensures that even small changes in data (such as operation time is more than 1 second) will result in a completely different hash value, the hash value will be bound to the corresponding learning content (such as "physics - Newton's second law"), forming a "content-data-hash" ternary association, which is convenient for accurate tracing;

[0090] And the blockchain uses a consortium chain architecture (such as Hyperledger Fabric), which is maintained by schools, teachers and technical parties. Each piece of data needs to be confirmed by more than 3 nodes to ensure that the data cannot be tampered with, and can effectively prevent malicious modification of learning records (such as error rate data of virtual experiments).

[0091] Access management module: set data access permission level by user, support authorized object to trace corresponding learning track original data through hash value, ensure data authenticity and controllability;

[0092] Among them, the permission level is divided into three levels: public (everyone can see), teachers can see, private, users can set permissions through role-based access control (RBAC) model, for example "allow math teachers to view physics experiment data of students in this class";

[0093] And authorized access needs to be verified through digital signature, when authorized object requests access, it needs to submit an application containing its public key, and verify the validity of the signature through RSA algorithm (signature verification is passed and permission is matched, access is allowed).

[0094] Of course, the above description is not a limitation of the present application, and the present application is not limited to the above examples. Changes, modifications, additions or substitutions made by ordinary skilled in the art within the scope of the present application also belong to the protection scope of the present application.

Claims

1. An interactive education platform based on virtual reality technology, characterized in that, Comprise: The acquisition and preprocessing module is used for acquiring multi-source interactive data and cleaning and standardizing, providing high-quality basic data input for the system, guaranteeing the reliability of subsequent module analysis, solving the problem of poor data quality and non-uniform format affecting interactive effect in traditional technology; The scene generation and interaction module is used for constructing a dynamic virtual scene and realizing multi-modal interactive feedback, providing an immersive interactive environment for users, solving the problem of static scene and single interaction mode in traditional technology; The adaptive learning module is used for collecting user interaction data, analyzing learning state and dynamically adjusting scene content, realizing personalized learning adaptation, and solving the problem of fixed content and lack of targeted adjustment in traditional technology; The interaction module is used for capturing user expression features, calculating emotional state and driving virtual characters to respond, enhancing emotional resonance, and solving the problem of lack of emotional interaction in the interactive process; The evaluation module is used for collecting learning data from behavior, cognition and emotion, and then converting it into radar chart and fluctuation curve through visual processing, and real-time feedback of learning situation, solving the problem of single evaluation dimension and feedback lag in traditional evaluation, providing basis for bidirectional optimization of teaching and learning; The storage module generates an unforgeable hash value storage for virtual project achievements and key learning nodes of innovative operations, and at the same time, builds a permission management mechanism, which is controlled by users to control data access permission, supports authorized tracking of learning track, solves the problem of easy tampering of learning data and difficult authentication, and provides a reliable basis for admission and ability authentication.

2. The interactive education platform based on virtual reality technology according to claim 1, characterized in that: The acquisition and preprocessing module comprises an acquisition module and a preprocessing module. The acquisition module: through VR head-mounted sensor, handle sensor and camera, user physiological data, operation behavior data and environmental interaction data are collected; The preprocessing module: filtering noise in collected data, standardizing effective data format, and outputting structured data which can be directly used for analysis of subsequent modules. 3.The interactive education platform based on virtual reality technology of claim 1, wherein: The scene generation and interaction module comprises a scene generation module and a feedback module. The scene generation module: based on UE5 engine combined with programmatic generation algorithm, according to teaching target, model library is called to dynamically build interactive virtual scene, supporting user to disassemble, combine and adjust parameters of scene elements; The feedback module: integrating multi-modal feedback logic, including eye tracking, gesture recognition and tactile feedback, real-time feedback is given to enhance the immersion and guidance of interaction.

4. The interactive education platform based on virtual reality technology according to claim 1, characterized in that: The adaptive learning module comprises an interactive collection module, an analysis module and an adjustment module. The interactive collection module: real-time collection of user interaction data in virtual scene, including operation time, error type, path selection and repeated operation times; The analysis module: using LSTM neural network to model interactive data, identifying user learning state label, and outputting state feature vector; The adjustment module: dynamically adjusting scene content according to learning state feature vector.

5. The interactive education platform based on virtual reality technology according to claim 1, characterized in that: The interaction module comprises an expression capture module, a response module and an emotion calculation module. The expression capture module: through camera, user facial muscle movement data is collected, key feature point coordinates of eyebrow arch and mouth corner are extracted, and expression feature sequence is generated; Response module: according to the dominant emotion type, the virtual character makes response, including expression simulation, voice adjustment and action output; Emotion calculation module: match the expression feature sequence with the preset emotion label library, calculate the emotion state confidence, and output the dominant emotion type.

6. The interactive education platform based on virtual reality technology according to claim 1, characterized in that: The evaluation module includes a data acquisition module and a visualization module. Data acquisition module: collect learning data from multiple dimensions, including behavior data, cognitive data, and emotional data; Visualization module: convert multi-dimensional data into intuitive charts, where radar chart is used to show the score distribution of behavior, cognition and emotion dimensions, and fluctuation curve is used to present the real-time change trend of each dimension data in the learning process.

7. The interactive education platform based on virtual reality technology according to claim 1, characterized in that: The evidence storage module includes a key node module and an access management module. Key node module: generate an unalterable hash value for the key learning nodes of virtual project achievements, innovation operations and evaluation results, and store them in the blockchain after associating with the time stamp; Access management module: set data access permission level by the user, support authorized objects to trace the original data of corresponding learning track through hash value, and ensure data authenticity and controllability.

Images